About Analyze My Shoulder
Who is behind Analyze My Shoulder
The site belongs to one independent operator — the team that develops JointLens, the umbrella product under which all of our imaging tools are built. We are a software team funded by the people who use the product, with no clinic, no laboratory and no equipment maker in the background. Anonymous medical content is a problem on the web; this page exists so that ours is not anonymous.
The reading the AI layer offers
Shoulder MRI, MR arthrography, CT and radiographs all open in the browser viewer, loose or zipped. After the series loads you scroll, adjust contrast, and decide which images are worth explaining. The models then write out what those rendered images contain in everyday words: the continuity and thickness of the supraspinatus and infraspinatus tendons, fluid in the subacromial-subdeltoid bursa, the shape of the acromion above the cuff, signal running into the superior labrum, bone loss on the anterior glenoid rim or an impaction defect on the humeral head, calcium deposits inside a tendon, and acromioclavicular joint changes.
Blind spots worth knowing about
The distinction between a high-grade partial tear and a small full-thickness defect is subtle even for experienced readers and is a frequent failure point here. Labral tears without contrast are easy to miss and easy to over-call. Adhesive capsulitis is largely a clinical diagnosis; imaging may look unremarkable in a shoulder that barely moves. Nothing in the output measures tendon retraction for surgical planning, grades fatty muscle atrophy, or knows whether your arm was internally or externally rotated in the scanner. No clinician reviews these paragraphs before you see them.
Guardrails around the analysis
The analyzer opens only after an explicit consent step describing the tool as informational, unvalidated and not a substitute for care. Reports repeat that framing, and our Medical Disclaimer documents the failure modes — fabricated findings, missed pathology, invented severity grades — in blunt terms. A shoulder that will not lift after trauma, sudden weakness, or a joint that looks out of place needs a clinician now, not an explanation later.
Where your images actually go
Nowhere, in the case of the original files. Decoding and display are done by code running on your device, so the raw study, the archive it came in and any PDF report you open stay inside the browser session. Requesting an explanation sends only rendered frames plus your typed context onward to the model providers; our servers pass that through without building a study archive behind it. Nothing persists for later retrieval — the mechanics are written out in the Privacy Policy.
Questions, complaints, corrections
All of it goes through one form rather than a published address. If a description of your cuff was wrong, say so — those reports are the main way the wording and the prompts improve.